handling securitythreats to the rfid system of epc networksgarcia_a/web/papers/... · used in rfid...

“K11012_C003.tex” — page 45[#1] 20/8/2010 10:31

Chapter 3

Handling Security Threatsto the RFID System of EPCNetworks

Joaquin Garcia-Alfaro, Michel Barbeau, and Evangelos Kranakis

Contents3.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463.2 Threat Analysis Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.3 Evaluation of Threats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.3.1 Authenticity Threats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483.3.2 Integrity and Availability Threats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.4 Survey of RFID Security Defences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533.4.1 Hardware-Based Primitives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533.4.2 Software Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

3.5 Future Directions for Research. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Terminologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Questions and Sample Answers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Author’s Biography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

3.1 IntroductionPassive radio frequency identification (RFID) is a wireless communication technology that allowsthe automatic identification of objects, animals, and persons through radio waves. Passive RFID tagsare electronic labels without self-power supply. They are energized by the electromagnetic field ofradio frequency (RF) front-end devices (hereinafter referred as RFID readers). The radio spectrum

45

“K11012_C003.tex” — page 46[#2] 20/8/2010 10:31

46 � Security of Self-Organizing Networks

used in RFID systems varies from low-frequency (LF) and high-frequency (HF) bands (typically125 kHz and 13.56 MHz) to ultra-high-frequency (UHF) bands (typically 868 MHz in Europe,915 MHz in North America, and 950 MHz in Japan). Distances from which the RFID tags can beinterrogated vary with the frequency band. It may vary from a few centimeters, while using LF andHF, to a few meters, while using UHF.

Although no single technology is ideal for all applications [1], most of the modern RFID systemsseem to be moving toward increasing the integration of long-distance passive tags into self-organizingwireless applications. This is the case with the modern electronic product code (EPC) Gen2 tags.They are becoming truly pervasive in wireless network applications, such as Mobile Wireless AdHoc Networks (MANETs), Wireless Sensor Networks (WSNs), and Vehicular Ad Hoc Networks(VANETs) [2]. Tags are potentially the targets of attack against their security and this raises majorconcerns. The objective of this chapter is to analyze some of these concerns and survey solutionsthat handle them.

3.1.1 BackgroundThe EPC technology originates from the MIT’s Auto-ID Center (now called the Auto-ID Labs). Ithad been further developed by different working groups at EPCglobal Inc. [3]. It is a layered service-oriented architecture to link objects, information, and organizations via Internet technologies. Atthe lowest layer, an identification system based on passive RFID tags and readers provides the meansto access and identify objects in motion. This system possesses two primary interfaces: the Class 1Generation 2 UHF Air Interface Protocol Standard (Gen2 for short) and Low Level Reader Protocol(LLRP).The former defines the physical and logical requirements for RFID readers (or interrogators)and passive tags (or labels).The latter specifies the air interface and interactions between its instances.

The next layer consists of a middleware composed of several services (such as filtering, fusion,aggregation, and correlation of events) that perform real-time processing of tag event data and collectthe identifier of objects interrogated by RFID readers at different time points and locations. Datagathered by sensors, such as temperature and humidity, can also be aggregated at the middlewarelayer within tag events. The middleware forwards the complete set of events to a local repositorywhere they are persistently stored (e.g., into a relational or XML database). The Reader Protocol(RP) and Reader Management (RM) interfaces define the interactions between a device capable ofreading/writing RFID tags and the middleware. The middleware relies on a second interface calledApplication Level Event (ALE) for interaction with other applications (e.g., repository managers).At the top of the architecture, the EPC Information Services (EPCISs) offer the means to accessthe data stored in EPC network repositories. These EPCISs are implemented using standard Webtechnologies such as the Simple Object Access Protocol (SOAP) and Web Services DescriptionLanguage (WSDL). Two additional services are defined for accessing the EPCIS of a given EPCnetwork by external applications: a lookup service binding object identifiers and EPCISs, calledthe Object Name Service (ONS); and a EPC discovery service (EPCDS) to perform searches withhigh-level semantics (i.e., similar to Web engines for Web page browsing).

Security attacks can target the different services of the EPC network architecture. They maysucceed if weaknesses within the underlying technologies are not handled properly. The exchangeof information between EPC tags and readers, for example, is carried out via wireless channelsthat do not posses basic security attributes such as authenticity, integrity, and availability. Thissituation allows attackers to misuse the RFID service of an EPC network and perform unauthorizedactivities such as eavesdropping, rogue scanning, cloning, location tracking, and tampering of data.The attacker motivation for performing these activities is potentially high. The attacker can obtain

“K11012_C003.tex” — page 47[#3] 20/8/2010 10:31

Handling Security Threats to the RFID System of EPC Networks � 47

financial gains (e.g., offering services for corporate espionage purposes). Mechanisms at the RFIDlevel of the EPC architecture must be applied to mitigate these security risks.

The implementation of new security features in EPC tags faces several challenges, the mainone being cost. The total cost of an EPC tag was estimated in [4] to be less than 10 cents perunit. The goal is to maintain a low cost. Other challenges include compatibility regulations, powerconsumption, and performance requirements [5]. In this chapter, we analyze threats to the securityof the exchange of information between RFID readers and tags. Some of them need to be han-dled by appropriate countermeasures. Our threat analysis is based on a methodology proposed bythe European Telecommunications Standards Institute (ETSI). It proposes the ranking of threatsdepending on their likelihood of occurrence, their possible impact on targeted systems, and the riskthey represent [6] for corporate systems. The results of our analysis are intended for leading furtherresearch and developments of security of EPC-based technologies. We also study countermeasuresfor threats ranked at the critical or major level. We discuss the benefits and drawbacks associatedwith the surveyed solutions.

Section 3.2 outlines the methodology used for our analysis of threats. Section 3.3 presents theidentified threats and their risk assessments. Section 3.4 surveys traditional security defenses forRFID solutions. Section 3.5 discusses some directions and trends for further research.

3.2 Threat Analysis MethodologyWe define a threat as the objective of an attacker to violate security properties of a target system,such as authenticity, integrity, and availability. We define the attacker as an agent that is exploitinga vulnerability of the targeted system to carry out the threat. The exploitation of the vulnerabilityis defined as the attack. The security officer of the target system must put in place countermeasuresto reduce the risk of the undesirable activities associated with all the threats. Given the difficulty ofimplementing countermeasures for every possible threat against a system, it is crucial for securityofficers to identify threats with potentially high impact and insure the presence of countermeasures.This is indeed the objective of the threat analysis.

The methodology we use is based on a framework proposed by the ETSI [6]. ETSI identifiesthree levels of threats: critical, major, and minor. Each level depends on estimated values for thelikelihood of occurrence of the threat and its potential impact on a given system. The likelihoodof a threat (cf. Figure 3.1a) is determined by the motivation for an attacker to carry out an attack

None

Solvable

Diff

icul

tyLikelihood

Likely

Likely

Possible

UnlikelyMinor

Strong

(a) (b)

Unlikely Critical

MajorPossible

Low ModerateMotivation Impact

High LowMediumHigh

Figure 3.1 Likelihood and risk functions: (a) likelihood of a threat, (b) risk evaluation function.

“K11012_C003.tex” — page 48[#4] 20/8/2010 10:31


associated to the threat versus the technical difficulties that must be resolved by the attacker toeffectively implement the attack. The three levels of likelihood are: (1) likely, if the targeted systemis almost assured of being victimized, given a high attacker motivation (e.g., financial gains as a resultof selling private information or disrupting network services) and lack of technical difficulties (e.g.,a precedent for the attack already exists); (2) possible, if the motivation for the attacker is moderate(e.g., limited financial gains) and technical difficulties are potentially solvable (e.g., the requiredtheoretical and practical knowledge for implementing the attack is available); and (3) unlikely, incase there is little motivation for perpetrating the attack (e.g., few or no financial gains resulting fromthe attack) or if significant technical difficulties and obstacles must be overcome (e.g., theoreticalor practical elements for perpetrating the attack are still missing).

The impact of a threat evaluates the potential consequences on the system when the threat issuccessfully carried out. The following three categories are identified: (1) low, if the consequencesof the attack can be quickly repaired without suffering from financial losses; (2) medium, if theconsequences are limited in time but might result in few financial losses; and (3) high, if the attackresults in substantial financial loss and/or law violations. The risk of a threat is ranked in [6] as minor,if it is unlikely to happen and it has low or medium potential impact, or if it is possible but withlow potential impact. A threat is ranked as major if it is likely but has low potential impact, if it ispossible and has medium potential impact, or if it is unlikely but has high potential impact. A threatis ranked as critical if it is likely and has high or medium potential impact, or if it is possible andhas high potential impact. Through our experience with the ETSI methodology, we have observedthat several threats are overclassified as major, when they would better be ranked as minor. We haveslightly adapted the risk function in order to focus on truly critical or major threats. Figure 3.1bpresents the adapted risk function. A threat is ranked as major when its likelihood is possible and itspotential impact is medium. A threat is ranked as minor when it is unlikely to happen or when itspotential impact is low. Minor risk threats typically require no countermeasures. Major and criticalthreats need to be handled with appropriate countermeasures. Moreover, critical threats should beaddressed with the highest priority.

3.3 Evaluation of ThreatsThe communication channel between the components of the RFID system of an EPC network,that is, tags and readers, is a potentially insecure wireless channel. It is fair to assume that most ofthe threats on EPC configurations are going to target this level. We analyze threats targeting basicsecurity features such as authenticity, integrity, and availability during the exchange of data betweena RFID tag and a RFID reader. We assume that attackers may only act from outside when tryingto exploit the wireless channel between tags and readers, for example, the lack of authenticationbetween these elements. We therefore assume that attackers do have physical access neither to thecomponents of the system nor to the organization itself. The reason we ignore direct physical accessis because we assume the presence of other security mechanisms in the organization (e.g., physicalaccess control and surveillance of workers). Attackers, however, may have access to information aboutthe system and its components or services. We summarize in Table 3.1 the results of our evaluation.

3.3.1 Authenticity ThreatsThe EPC Gen2 standard is designed to balance cost and functionality [4]. However, security featureson board Gen2 tags are minimal. They protect message integrity via 16-bit Cyclic Redundancy

“K11012_C003.tex” — page 49[#5] 20/8/2010 10:31


Table 3.1 Evaluation of Threats

Threats Motivation Difficulty Likelihood Impact Risk

Eavesdropping, rogue scanning High Solvable Possible High Critical

Cloning of tags, location tracking Moderate Solvable Possible Medium Major

Tampering of data Moderate Solvable Possible High Critical

Destruction of data, denial of service Moderate Solvable Possible Medium Major

Malware Moderate Strong Unlikely Medium Minor

Codes (CRC) and generate 16-bit pseudorandom strings. Their memory, very limited, is separatedinto four independent blocks: reserved memory, EPC data, Tag Identification (TID), and usermemory. The absence of strong authentication on the tags opens the door to malicious readers thatcan impersonate legal readers and perform eavesdropping attacks. Figure 3.2 shows a simplifieddescription of the steps of the Gen2 protocol for product inventory. In Step 1, the reader queriesthe tag and selects one of the following options: select, inventory, or access [3]. Figure 3.2 representsthe execution of an inventory query. It assumes that a select operation has been completed in orderto single out a specific tag from the population of tags. When the tag receives the inventory query,it returns a 16-bit random string denoted as RN16 in Step 2. This random string is temporarilystored in the tag memory. The reader replies to the tag in Step 3 with a copy of the random string,as an acknowledgment. If the echoed string matches the copy of RN16 stored in the tag memory,the tag enters the acknowledged state and returns the EPC.

Let us observe that any compatible Gen2 reader can access the EPC. The traffic between tags andreaders flows through nonauthenticated wireless channels. Illegitimate collection of traffic might beslightly protected by reducing the transmission power or by sheltering the area. It is, although, theo-retically possible to conduct eavesdropping attacks. We define forward eavesdropping as the passivecollection of queries and commands sent from readers to tags; and backward eavesdropping as thepassive collection of responses sent from tags to readers. Although the range for backward eaves-dropping could be only of a few meters [3], and probably irrelevant for a real eavesdropping attack,the distance at which an attacker can eavesdrop the signal of an EPC reader can be much longer. In

Reader Tag

1. Query

2. RN16

3. ACK(RN16)

4. EPC

Figure 3.2 Inventory protocol of a Gen2 tag.

“K11012_C003.tex” — page 50[#6] 20/8/2010 10:31


ideal conditions, for example, readers configured to transmit at maximum output power, the signalcould be received from tens of kilometers away. Analysis attacks inferring sensitive informationfrom forward eavesdropping, for example, analysis of the pseudorandom sequences generated bythe tags (denoted as RN16 in Figure 3.2), are hence possible. Replay attacks enabled by this inferreddata are also possible. The absence of a strong authentication process also enables scanning attacks.Although, the distance at which an attacker can perform scanning is considerably shorter than thedistance for forward eavesdropping. The use of special hardware (e.g., highly sensitive receivers andhigh gain antennas) could enable rogue scanning attacks.

We can conclude that outsiders equipped with Gen2 compatible readers and special hard-ware can theoretically eavesdrop the communication between readers and tags; or scan objects inmotion if they successfully manage to place their readers at appropriate distances. According to[3], the information stored on an EPC tag is limited to an identification number. No additionaldata beyond the number itself is conveyed in the EPC. Additional information associated withthe code must be retrieved from an EPCIS. However, an attacker accessing these data may deter-mine types and quantities of items in a supply chain and sell the information to competitors orthieves. An attacker can obtain information from the EPC, that is, the manufacturer and prod-uct number. This information may be used for corporate espionage purposes by competitors, orfor other attacks against other services of the EPC infrastructure. Clearly, the motivation of anattacker to carry out this threat must be rated as high, since attackers can sell their services tocompetitors, thieves, or any other individual looking for the objects tagged in the organization.The difficulties for performing both eavesdropping and rogue scanning, as shown by the exam-ple depicted by Figure 3.2, are solvable. This level of motivation and degree of difficulty lead toa likelihood that is possible. Regarding the potential impact of these threats (e.g., disclosure ofinformation considered by the organization as confidential or trade secrets), it is high, since it mayhave serious consequences for an organization if an attacker offers the malicious service to com-petitors or to thieves. These threats are assessed as critical and need to be handled by appropriatecountermeasures.

Using the codes eavesdropped or scanned by unauthorized readers, an attacker may successfullyclone the tags by conducting, for example, skimming attacks. Indeed, an attacker can simply dumpdata and responses from a given tag, and program it into a different device. The objective of theattacker for performing the cloning of tags is the possibility for counterfeiting. The attacker maycreate fake EPC tags that contain data and responses of real tags and sell these counterfeit tagsfor profit. The forgery of legal tags can be performed without physical access to the organization.We rank the motivation of attackers to carry out the attacks associated with this threat as moderatesince they can obtain some financial gain by offering this service to third parties. Current EPC spec-ifications do not include any mechanism for Gen2 compatible readers to verify if they communicatewith genuine or fraudulent tags. We thus rate the difficulties associated to this threat as solvable.This level of motivation and degree of difficulty lead to a likelihood that is possible. Regarding thepotential impact of this threat, it is medium and thus the threat is assessed as major.

The lack of a strong authentication process in Gen2 tags also has consequences to the privacyof tagged object bearers. Indeed, interrogations of Gen2 tags give attackers unique opportunitiesfor the collection of personal information (and without the consent of the bearer). This can haveserious consequences, such as location tracking or surveillance of the object bearers. An attackercan distinguish any given tag by just taking into account the EPC number. Following a reasoningsimilar to the one used for the cloning threat leads to ranking the risk of the location track-ing threat as major. This threat, as well as the cloning threat, must be handled by appropriatecountermeasures.

“K11012_C003.tex” — page 51[#7] 20/8/2010 10:31


3.3.2 Integrity and Availability ThreatsGen2 tags are required to be writable [3]. They must also implement an access control routine,based on the use of 32-bit passwords, to protect the tags from unauthorized activation of thewriting process. Other operations, also protected by 32-bit passwords, can be used in order topermanently lock or disable this operation. Although the writable feature of Gen2 tags is veryinteresting, it is also one of the least exploited features in current EPC scenarios (due probably tothe lack of a strong authentication process, as reported in the previous section). Writable tags arehence locked in most of today’s EPC applications.This option will, however, be extremely importantin future EPC applications, especially on those self-organizing-based scenarios, where the additionof complementary information into the memory of the tags will require the unlocking of the writingprocess (e.g., to store routing parameters, locations, or time stamps). It is therefore important toanalyze the risk of a tampering attack to the data stored by Gen2 tags, if they can be accessedin write mode from a wireless channel that does not guarantee strong authentication. Figure 3.3presents a simplified description of protocol steps for requesting and accessing the writing processthat modifies the memory of a Gen2 tag. We assume that a select operation has been completed, inorder to single out a specific tag from the population of tags. It is also assumed that an inventoryquery has been completed and that the reader has a valid RN16 identifier (cf. Figure 3.2, Steps 2 and3) to communicate and request further operations from the tag. Using this random sequence (cf.Figure 3.2, Step 5), the reader requests a new descriptor (denoted as Handle in the following steps).This descriptor is a new random sequence of 16 bits that is used by the reader and tag. Indeed,any command requested by the reader must include this random sequence as a parameter in thecommand. All the acknowledgments sent by the tag to the reader must also include this randomsequence. Once the reader obtains the Handle descriptor in Step 6, it acknowledges by sendingit back to the tag as a parameter of its query (cf. Step 7). To request the execution of the writingprocess, the reader needs first to be granted access by supplying the 32-bit password that protects the

5. Req_RN (RN16)

7. Req_RN (Handle)

8. RN16′

9. Access (Pin31:16⊕RN16′)

6. Handle

10. Handle

14. Handle

16. Header, Handle

15. Write (membank, wordptr,data, handle)

11. Req_RN (Handle)

12. RN16″

Reader Tag

13. Access (PIN15:0⊕RN16″)

Figure 3.3 Writing protocol of a Gen2 tag.

“K11012_C003.tex” — page 52[#8] 20/8/2010 10:31


writing routine. This password is actually composed of two 16-bit sequences, denoted in Figure 3.3as PIN31:16 and PIN15:0. To protect the communication of the password, the reader obtains in Steps8 and 12, two random sequences of 16 bits, denoted in as RN16 ′ and RN16 ′′. These two randomsequences RN16 ′ and RN16 ′′ are used by the reader to blind the communication of the passwordtoward the tag. In Step 9, the reader blinds the first 16 bits of the password by applying an XORoperation (denoted by the symbol ⊕ in Figure 3.3) with the sequence RN16 ′. It sends the result tothe tag, which acknowledges the reception in Step 10. Similarly, the reader blinds the remaining 16bits of the password by applying an XOR operation with the sequence RN16 ′′, and sends the resultto the tag in Step 13. The tag acknowledges the reception in Step 14 by sending a new Handle to thereader. By using the latter, the reader requests the writing operation in Step 15, which is executedand acknowledged by the tag in Step 16.

An attacker can find the 32-bit password that protects the writing routine. It suffices to interceptsequences RN16′ and RN16′′, in Steps 8 and 12, and to apply the XOR operation to the contentsof Steps 9 and 13. Other techniques to retrieve similar passwords have also been reported in theliterature. For example, in [7] the authors present a mechanism to retrieve passwords by simplyanalyzing the radio signals sent from readers to tags. Although the proof-of-concept implementationof this technique is only available for Gen1 tags [3], the authors state that Gen2 tags are equallyvulnerable. The technical difficulties for setting up attacks to retrieve the password are thereforeranked as solvable. The likelihood of the tampering threat is classified as possible. Regarding theimpact, it is ranked, because of some extreme scenarios, as high. For example, in the context of apharmaceutical supply chain, corrupting data in the memory of EPC tags can be very dangerous:the supply of medicines with wrong information, or delivered to the wrong patients, can lead tosituations where a sick person could take the wrong drugs. In these circumstances, the combinationof likelihood and impact of the tampering threat lead to critical risk. The threat needs therefore tobe addressed by appropriate countermeasures.

Let us note that these attacks enabled by retrieving the passwords, that protect both writing andself-destruction routines of Gen2 tags [3], can be used as models to analyze the risk of threats likedestruction of data or denial of service [8]. Tag information can also be destroyed by devices thatsend strong electromagnetic pulses. Devices, such as the RFID-zapper [9], have been presented inthe literature. We can also include here denial of service attacks consisting of jamming channels orflooding channels between tags and readers by sending a large number of requests and responses. Forperforming a jamming attack, the attacker uses powerful transmitters to generate noise in the rangeof frequencies used by readers and tags. In any case, the technical difficulties are ranked as solvable.The motivation of attackers to carry out these threats is rated as moderate, since they can obtainsome financial gain by offering their malicious services. The likelihood of these two threats is henceclassified as possible. However, since the impact of these threats represents to the victim temporaldisruption of its operations rather than great financial losses, we rate the impact as medium, and sothe risk of these two threats as major.

The final threat analyzed in this section, related to attacks to the integrity and availability of theback-end servers connected to RFID readers, was initially reported in [10]. Rieback et al. uncover thepossibility of using malware to attack back-end databases.Their approach classifies such malware intothree categories: (1) exploits, (2) worms, and (3) viruses. The exploits are attacks carried out withinthe information stored into RFID tags. They target the security of middleware services connectingreaders to back-end databases. Worms and viruses are attacks that spread themselves over new RFIDtags by using network connections (in the case of worms) or connectionless self-replication strategies(in the case of viruses). The malware reported in [10] exploits the trust relationship between back-end databases and the information sent by readers—obtained in turn from malicious tags. Rieback

“K11012_C003.tex” — page 53[#9] 20/8/2010 10:31


et al. consider that even if there is a very tiny window for storing information into an RFID tag,traditional attacks against information systems (e.g., buffer overflows and SQL injection attacks)might be condensed into a small string of bits harmful enough to break the security of a system. Theauthors present a proof-of-concept that uses tags carrying an SQL injection attack that compromisesthe security of the back-end layer of an ad hoc RFID setup. The work presented in [10] is interestingand relevant. However, we think that the likelihood for those threats must be rated as unlikely, sinceno real-world vulnerabilities on the filtering and collection of middleware services specified byEPCglobal can be exploited at the moment. We conclude that even if the impact is potentiallyserious, due to its unlikely degree of likelihood, the threat is assessed as minor.

3.4 Survey of RFID Security DefencesResearch on security countermeasures for RFID technologies can be divided into two categories: (1)hardware-based security primitives for RFID tags, and (2) software protocols using the hardware-based primitives. We review in this section a nonexhaustive list of contributions in both categories.

3.4.1 Hardware-Based PrimitivesAccording to research presented in [4], the cost of passive EPC tags should not exceed 5 cents tosuccessfully enable their deployment on worldwide scale. Of these 5 cents, only 1 or 2 should be usedfor the manufacturing of the integrated circuit (IC). Another challenge is that the available layoutarea for the implementation of the IC is in the order of 0.25 mm2 which, considering current com-plementary metal–oxide–semiconductor (CMOS) technology, corresponds to a theoretical numberof logic gates from 2000 to 4000. Not all the barriers identified in [4] have been removed. Today,the EPC technology is more expensive than what it was originally anticipated—around 10 centsper unit in large quantities. The inclusion of additional features, especially for security purposes,may increase the total end-cost of tags up to 15 cents per unit or more. Although Moore’s Law saysthat the cost of ICs will continue to decrease, cost of analogue devices (i.e., RF front-end of tags)is relatively stable and will remain a constraint [11]. The inclusion of new elements must thereforebe clearly justified.

Since EPC tags are powered from the weak energy captured from a reader’s electromagneticwaves, their current consumption also needs to be taken into account. This consumption variesaccording to the operation that is being performed [12] (e.g., responding to a query or writingdata into the memory) and other parameters such as the transmission rate, response delay, andmemory technology. Most of the operations performed within the modern EPC tags consumeabout 5–10 μamps—although some special operations, such as write accesses, may consume more.The current consumption of new security primitives must be within this range to allow low-costtag production. They must also work at the data rate of EPC applications. For example, somesupply chain applications demand an average reading speed of about 200 tags/s. This leads to a datatransmission rate from tag to reader, of about 640 kbps; and from reader to tag of about 120 kbps.Delays associated to new security mechanisms (e.g., time to perform encryption or random numbergeneration) may also affect the global performance. Delays must hence be taken into account andminimized. We refer the reader to [11] for a more detailed description of aspects that must beconsidered during the design of new primitives.

Several security proposals aim at including cryptographic primitives on low-cost EPC tags. How-ever, not all the proposals meet the aforementioned constraints or guarantee secure designs. Existing

“K11012_C003.tex” — page 54[#10] 20/8/2010 10:31


implementations of one-way hash functions, such as MD4, MD5, and SHA-128/SHA-256, exceedcost constraints due to the required number of gates—from 7000 to over 10,000 logic gates accord-ing to [13]. The use of cellular automata (CA) theory for the implementation of one-way functions[14] and encryption engines [15]—typically built upon feedback shift registers with a much lowernumber of gates—has been investigated for the implementation of cryptographic primitives onlow-cost RFID tags. However, it has been proved that these implementations are insecure [16,17].Similarly, the use of linear feedback shift registers and nonlinear feedback shift registers (LFSR &NLFSR) as underlying mechanisms for the implementation of low-cost one-way hash functionsand pseudorandom number generators (PRNG)—without appropriate measures that increase thecost—also lead to insecure implementations [18,19]. Light-weight hardware implementations ofstandard block ciphers to implement one-way hash functions have been discussed. The use of ellip-tic curve cryptosystems (ECC) [19] for the implementation of primitives for RFID tags has beendiscussed in [20]. Its use of small key sizes is seen as very promising for providing an adequatelevel of computational security at a relatively low cost [11]. An ECC implementation for low-costRFID tags can be found in [21]. In [22], on the other hand, Feldhofer et al. present a 128-bitimplementation of the advanced encryption standard (AES) [23] on an IC of about 3500 gateswith a current consumption of less than 9 μamps at a frequency of 100 kHz. The encryption ofeach block of 128 bits requires about 1000 clock cycles. Although, it considerably simplifies previ-ous implementations of the AES, for example, proposals presented in [24,25] that require between10,000 and over 100,000 gates, respectively, the design is still considered too complex for basicEPC setups [26].

More suitable encryption engine implementations can be found in [27,28]. The first referencepresents the implementation of the tiny encryption algorithm (TEA) [29]. It is implemented onan IC of about 3000 gates with a current consumption of about 7 μamps. It fits the timingrequirements of basic EPC setups where hundreds of tags must simultaneously be accessed by thesame reader. For meeting the constraints, the implementation relies on very simple operators suchas XOR, ADD, and SHIFT. The authors of TEA [29] claim that, despite its simplicity and ease ofimplementation, the complexity of the algorithm is equivalent to data encryption standard (DES)[19]. Variants of the basic algorithm, such as eXtended TEA (XTEA), are however necessary forimplementing one-way hash functions. Mace et al. discuss in [28] some of the vulnerabilities ofTEA, such as linear and differential cryptanalysis attacks, and present scalable encryption algorithm(SEA). Given the relatively recent invention of these algorithms, their strength is not clear [11].

There are other hardware-based security enhancements for RFID technologies not relying onthe implementation of cryptographic primitives. Many signal- and power-based defenses, such asshielding of tags, use of noise and third-party blocker devices have been surveyed in [26]. The useof distance measurements to detect rogue readers has been discussed. In [30], for example, Fishkinet al. propose the inclusion of low-cost circuitry on tags to use the signal-to-noise ratio of readers as ametric for trust. In [31], a similar assumption is used in order to claim that a reader can be authorizedto read a tag contents according to its physical distance. The use of trust [32] and trusted computing[33] with similar purposes has also been discussed. For example, Molnar et al. describe in [33] a mech-anism consisting of trusted platform modules (TPMs) to enforce privacy policies within the RFIDtags. A trusted entity called trusted center (TC) decides whether readers are allowed or not to accesstags. Finally, the use of radio fingerprinting [34,35] to detect characteristic properties of transmittedsignals and design authentication procedures has been investigated. The authors in [11] consider,however, that this technique is difficult to develop on RFID applications and that the benefits ofusing it, with respect to performance, cost, and required implementation surface on tags, are unclear.Avoine and Oechslin also debate in [36] the prevention of the traceability via radio fingerprinting.

“K11012_C003.tex” — page 55[#11] 20/8/2010 10:31


They conclude that obtaining radio fingerprint of tags is expensive and difficult. The myriad of tagsin circulation in future RFID scenarios makes impracticable the individual distinction of them.

Physically unclonable functions (PUFs) and physical obfuscated keys (POKs) are promising forthe implementation of new security primitives in low-cost EPC tags. They can be used to handlethe authentication threat, as well as the cloning and location tracking threats. Half way betweencryptography-based enhancements and physical protection defenses, the ideas behind PUFs andPOKs originated in [37] with the conception of optical mechanisms for the construction of physicalone-way functions (POWFs).Their use to securely store unique secret keys, in the form of fabricationvariations, was proposed as a silicon prototype in [38,39]. These ideas were later improved in [40].A coating PUF proposed in [41] is implemented with less than 1000 gates. These designs exploit therandom variations in delays of wires and logic gates of an IC. For example, the silicon PUF presentedin [39] receives input data, as a challenge, and launches a race condition within the IC: two signalspropagate along different paths and are compared to determine which one comes first. To decidewhich signal comes first, a controller, implemented as a latch, produces a binary value. Holcombet al. [42] propose using the SRAM based on CMOS circuitry to generate physical fingerprints. Thekey idea is the use of SRAM start-up values as origin of randomness. The use of 256 bytes of StaticRandom Access Memory (SRAM) can yield 100 bits of true randomness each time that the memoryis powered up. While sound in theory, this technique has as important drawback the limitation ofmemory space of current low-cost tags. The implementation of PUF-based circuits seems to haveclear advantages at a cost of less than 1000 logic gates [41]. This technology provides a cost effectiveand reliable solution that meets the constraints and requirements. Drawbacks, such as the effectsof environmental conditions and of power supply voltage [43], must be taken into account. Thedifficulty of successfully modeling the circuits and their reliability have also raised some concerns.Bolotnyy and Robins [44] address some of these issues. Some attacks on PUF- and POK-basedprotocols are outlined in [40]. The execution and reinterpretation of existing protocols via newPUF and POK designs—essentially the challenge–response protocols—are outlined in the sequel.

3.4.2 Software ProtocolsWe review algorithmic solutions and software protocols for handling the threats uncovered in Section3.3. The solutions rely on the implementation and use of hardware-based primitives discussed inSection 4.1.

Message Authentication Code (MAC)-based security protocols for wireless applications is atypical solution discussed in the literature (e.g., [44–46]). In [45], Takaragi et al. present a verysimple MAC-based approach. It uses a static unrewritable 128-bit identifier stored, at manufacturingtime, in every tag. This static identifier is not modifiable once the shipment is made. To build upthis identifier, the manufacturer uses a unique secret key for each tag and a keyed hash functionthat accepts as input the secret key and a specific message. All this information (i.e., secret key, hashfunction, and specific message) is communicated by the manufacturer to the client. By sharing thisinformation among readers and tags, integrity and authenticity of exchanged messages is verified.It therefore reduces the risk of threats to authenticity and integrity by increasing the technicaldifficulties of performing attacks. However, due to the use of static identifiers embedded in the tagsat manufacturing time, the location tracking issue is not solved. Moreover, brute force attacks canbreak the secrets shared between readers and tags.

The use of public key cryptography and digital signatures is discussed in [47]. The authorsaddress the protection of banknotes embedding the RFID tags. Their approach includes the pos-sibility of deploying cryptographic protocols in RFID applications, but avoids the need to embed

“K11012_C003.tex” — page 56[#12] 20/8/2010 10:31


cryptographic primitives within the tags. The scheme consists of a public-key cryptosystem used bya central bank aiming to avoid banknote forgery and a law enforcement agency that aims at trackingbanknotes. Both authorities, that is, central bank and law enforcement agency, hold an independentpair of public and private keys associated to each banknote. The central bank authority assigns aunique serial number to each banknote. The central bank authority, using its private key, signs theunique serial number. The unique serial number of the banknote and its corresponding digital sig-nature are printed on the banknote as optical data. In addition, the law enforcement agency encryptswith its public key the digital signature, unique serial number, and a random number. The resultingciphertext is stored into a memory cell of the RFID tag. This memory cell is keyed-protected. Thetag only grants write access to this memory cell if it receives an access key derived from the opticaldata. The random number used to create the ciphertext is also stored into a separated memory cellof the tag. This second memory cell is also keyed-protected. The tag only grants read or write accessto this memory cell if it receives an access key derived from the optical data.

Now, a merchant that receives a banknote must verify first the digital signature, printed in thebanknote as optical data, using the public key of the central bank. Second, the merchant must alsoverify the validity of the ciphertext stored in the banknote’s tag. To do so, the merchant encrypts thedigital signature, serial number, and random number stored in the tag’s memory, using the publickey of the law enforcement agency and the optical data. If one of these two verification processesfails, the authorities must be warned. To avoid using the same ciphertext on every interaction, Juelsand Pappu propose the use of a reencryption process that can be performed by the merchant withoutthe necessity of accessing the private keys of the law enforcement authority. Indeed, based on thealgebraic properties of the El Gamal cryptosystem [19], the initial ciphertext can be transformed intoa new unlinkable ciphertext only using the public key of the law enforcement authority [26]. Thisreencryption process is performed outside the tags. Integrity issues of this approach are discussedand fixed in [48]. However, the whole process and requirements for implementing the approach in[47,48] are too complex and expensive for use in EPC supply chain applications.

Mutual authentication protocols among tags and readers are discussed in [49,50]. The workpresented by Kinosita et al. in [49] consists of an anonymous ID scheme, in which a tag containsonly a pseudonym that is periodically rewritten. Pseudonyms are used instead of real identifiers (e.g.,instead of the EPC codes). Similarly, the approach of Juels entitled minimalist cryptography forlow-cost RFID tags [50] suggests a very lightweight protocol for mutual authentication between tagsand readers based on one-time authenticators. Both solutions rely on the use of pseudonyms andkeys stored within tags and back-end servers. Each tag contains a small collection of pseudonyms,according to the available memory of the tag. A throttling process is used to rotate the pseudonyms.Each time the tag is interrogated by a reader, a different pseudonym is used in the response.Authorized readers have access to the complete list of pseudonyms set for each tag and can correlatethe responses they receive. Without the knowledge of this list, unauthorized readers are unable toinfer any information about the several occurrences of the same tag. The process also forces tags toslow down their data transmissions when queried too frequently, as a defense to potential brute-force attacks. The memory space on current low-cost tags is the main limitation of this approach.Although enhancements can be used to update the list of pseudonyms, communication costs, andintegrity threats will remain as main drawbacks.

The use of hash-lock schemes for addressing authentication issues is another possibility. A designcan be found in [51]. Weis et al. propose a way to lock tags without storing access keys in them.Only hashes of keys must be known by the tags. Keys must be also stored on back-end servers andbe accessible by authorized readers. Most authentication threats are therefore mitigated by lockingtags. Cloning and tracking threats are handled by avoiding the use of real identifiers once tags are

“K11012_C003.tex” — page 57[#13] 20/8/2010 10:31


locked. In [52], Henrici and Müller extend the hash-lock scheme and address some weaknessesin [51] to increase traceability and location resistance. A similar hash-based protocol is presentedin [53] in order to deal with those limitations by using time stamps. Other similar hash-basedprotocols for handling authentication threats can be found in [54–56]. All these protocols rely onsynchronized secrets residing in the tags and back-end servers. They require a one-way hash functionimplemented within the tags. The requirement of reliable hash primitives implemented at the taglevel is the main drawback. Workload on back-end servers is also considerably high and can makedifficult the deployment in real-world EPC supply chain applications. The Yet Another Trivial RFIDAuthentication Protocol (YATRAP) protocol presented in [57] reduces the cost of computationby combining precomputed hash-tables for tag verification processes, use of time stamps, andgeneration of pseudorandom numbers. The protocol is, however, vulnerable to availability attackswhen temporal de-synchronizations between tags and readers occur. Some limitations are addressedin [58]. Chatmon et al. define new protocols for anonymous authentication. These improvementsnotably increase the degree of workload on servers and are highly complex for use in supply chainapplications.

3.5 Future Directions for ResearchAlgorithmic solutions avoiding the execution of on-tag cryptographic processes seems to lead thefuture of research in RFID security. In this sense, a secret-sharing scheme is presented by Juels et al.in [59] as a defense against the authenticity threats in EPC supply chain applications. Two differentmodels are discussed: dispersion of secrets across space and dispersion of secrets across time. Bothmodels are based on a secret-sharing strategy, where a secret used to encrypt EPCs is split in multipleshares and distributed among multiple parties. In order to obtain the EPC of a tag, a party mustcollect a minimum number of shares distributed among all the other parties. Authentication istherefore achieved though the dispersion of secrets. The dispersion helps to improve the authen-tication process between readers and tags, as tags move through a supply chain. Assuming that agiven number of shares is necessary for readers to obtain the EPCs assigned to a pallet, for example,a situation where the number of shares obtained by readers is not sufficient to reach the thresholdprotects the tags from unauthorized scanning (i.e., unauthorized readers that cannot obtain thesufficient number of shares cannot obtain the EPCs either). The approach can be implemented onEPC Gen2 tags without requiring any change to the current tag specification. A limitation is theamount of tag memory space required for storing the shares. However, the shrinking of shares canallow the application of the scheme to current EPC tags. A more important problem is that thelocation tracking threat is not addressed. Indeed, the shares used in the approach are static. Thisproblem must be solved before deployment of the scheme.

Challenge–response protocols for low-cost EPC tags using physical unclonable functions (PUFs)and POKs have recently gained importance. An approach presented in [60], based on PUFs proposedin [38,39], consists of a challenge–response scheme that probabilistically ensures unique identifica-tion of RFID tags. A back-end system must learn challenge–response pairs for each PUF/tag. It thenuses these challenges (hundreds of them) at a time, to identify and authenticate tags. Unique iden-tification of tags is only ensured probabilistically. The exposition of tag identifiers to eavesdroppersand lack of randomness in tag responses, make the approach vulnerable to the location trackingthreat. Moreover, the great number of challenges that are necessary in the identification processincreases the tag response delay and power consumption. Hence, this approach might not meetthe constraints and requirements mentioned in Section 4.1. An alternative approach is presented

“K11012_C003.tex” — page 58[#14] 20/8/2010 10:31


in [61]. Tuyls and Batina discuss an off-line PUF-based mechanism for verifying the authenticityof tags through the PUF technology presented in [41]. Similar to the results presented in [50,62],where readers and tags define ad hoc secrets, the PUF-based approach uses the internal physicalstructure of tags to generate unique keys. A key extraction algorithm from noisy binary data is pre-sented in [61]. The usage of PUF-based keys simplifies the process of verifying tag authenticity. Thecombination of unique keys generated onboard together with the use of signatures avoid leakingof a single identifier and increases the technical difficulties for an attacker to carry out the locationtracking threat. The main drawback is the need of large storage space and reliable searching processeson back-end servers to link readers with PUF/tag identifiers. The use of public key and digital sig-natures, based on Elliptic Curve Cryptography (ECC), is another important constraint. Followingthe trend of combining PUFs together with traditional cryptographic primitives and encryptionengines, a modification of the tree-based hash protocols proposed in [63] is presented in [64]. Usingthe notion of POKs introduced in [38] (i.e., application of a fixed hard-wired challenge to the PUFto obtain a unique secret), the authors guarantee the existence of internal keys in basic tree-basedhash protocols, now physically obfuscated. They cannot be cloned by unauthorized parties. The useof an AES engine, such as the one presented in [23], is proposed. On the other hand, Bolotnyy andRobins present in [44] a complete set of adapted MAC protocols, based on PUFs, trying to simplifythe challenge–response communication scheme of previous proposals and to eliminate requirementof traditional cryptographic primitives. Each tag generates multiple identifiers based on embeddedPUFs. Their approach only addresses static identification. It is vulnerable to the location trackingthreat identified in Section 3.3. It does not solve the requirement of huge lists of challenge–responsepairs for each PUF/tag that must be stored on back-end servers connected to the readers. Indeed,each given pair is of single use to prevent replay attacks.

3.6 ConclusionsAt the beginning of this chapter we presented an analysis of threats to the RFID system of the EPCarchitecture. We identified different groups of threats that we consider relevant for further research.We ranked the eavesdropping, rogue scanning, and tampering threats as critical; and cloning, track-ing, and denial of service threats as major. We concluded that they must be handled by appropriatecountermeasures. We then surveyed in the sequel practical and theoretical security defenses that canbe useful to reduce the risk of the identified threats. We looked at the different defenses from twodifferent research perspectives. On the one hand, we surveyed research on hardware-based defensesthat aim at providing additional security primitives on tags such as one-way hash functions, encryp-tion engines, and physically unclonable functions (PUFs). On the other hand, we surveyed researchon software protocols that make use of these new on-tag primitives for designing and implementingreliable algorithms for dealing with security and privacy issues. We have seen that the implementa-tion of well-known cryptographic primitives is possible and allows the design of software protocolsto reduce the risk of threats ranked as critical or major. The cost and requirements of these propos-als are the main difficulties. Indeed, they are too expensive for their deployment in supply chainscenarios based on the EPC technology. We have also surveyed the combination of cryptographicprimitives together with the use of PUFs for the design of cost-effective solutions. These solutionspresent drawbacks, such as the sensitivity of PUFs to physical noise and the difficulty to model andanalyze them. They are, however, promising solutions that successfully meet the implementationconstraints and requirements for handling the set of threats reported in our work. For the secondgroup, we conclude that the avoidance of on-tag cryptographic processes on current algorithmic

“K11012_C003.tex” — page 59[#15] 20/8/2010 10:31


solutions seems to lead the future directions of research in RFID security. In this sense, the useof secret-sharing schemes present clear advantages for the management of keys in the design ofauthentication protocols and to deal with privacy issues. The main drawback is the use of staticshares, limiting the use of this approach for addressing the location tracking threat.

AcknowledgmentsThe authors graciously acknowledge the financial support received from the following organizations:Natural Sciences and Engineering Research Council of Canada (NSERC), Mathematics of Infor-mation Technology and Complex Systems (MITACS), Spanish Ministry of Science and Education(TSI2007-65406-C03-03 “E-AEGIS” and CONSOLIDER CSD2007-00004 “ARES” grants), LaCaixa (Canada awards), and the IRF project LOCHNESS.

TerminologiesAdvanced Encryption Standard (AES)—A block cipher encryption standard, sponsored by the

National Institute of Standards and Technology (NIST), for protecting data.Countermeasure—A defense mechanism designed to mitigate the risk of a threat.EPC Network—A service-oriented architecture defined by EPCglobal Inc. that proposes the inte-

gration of RFID and Internet technologies to enable automatic identification and sharingof item data in supply chain applications.

Electronic Product Code (EPC)—Group of coding schemes defined by EPCglobal Inc.Elliptic Curve Cryptography (ECC)—A public key cryptosystem based on the algebraic structure

of elliptic curves over finite fields.European Telecommunications Standards Institute (ETSI)—Independent noncommercial orga-

nization that produces telecommunications standards to be used in Europe and beyond.EPC Number—A tag data format compatible with the family of coding schemes proposed by

EPCglobal Inc. It typically contains: a header, pointing out the family code that is beingused; a manufacturer code; an object class; and a serial number.

EPCGLOBAL Inc.—Joint venture between GS1 (Global Standards One, formerly known as EANInternational) and GS1 US™ (formerly the Uniform Code Council, Inc.) created tocommercialize the EPC technology.

Linear and Nonlinear Feedback Shift Registers (LFSR & NLFSR)—A digital circuit composed ofan n-bit shift register and a feedback function that generates pseudorandom sequences.

Passive Tag—RFID component attached to an item. It contains information about the item. Sinceit does not have its own power source, it provides the information by backscattering areader’s signal.

Physically Unclonable Function (PUF)—Hardware-based function embedded in a physical struc-ture, that is easy to evaluate but hard to reproduce.

Physical Obfuscated Keys (POK)—Hardware-based function for implementing secrets on digitaldevices using a physically unclonable function.

Pseudorandom Number Generator (PRNG)—Algorithmic solution to generate deterministicsequences of pseudorandom numbers.

Reader—RFID component that requests and receives information from tagged items.Tag Identification (TID)—Memory bank or identifier that uniquely identifies an RF tag.

“K11012_C003.tex” — page 60[#16] 20/8/2010 10:31


Threat Analysis—Determination and classification, in terms of importance, of threats targetingthe security of a system.

Tiny Encryption Algorithm (TEA)—A minimalist block cipher encryption algorithm for protect-ing data.

Trusted Platform Modules (TPMs)—Hardware-based cryptographic mechanism installed on themotherboard of a digital device (i.e., a personal computer) to enforce security protectionand trustworthiness.

Scalable Encryption Algorithm (SEA)—A block cipher encryption algorithm designed to be usedon embedded applications such as microcontrollers.

Questions and Sample Answers1. What is Passive RFID?

Passive radio frequency identification (RFID) is a wireless communication technology thatallows the automatic identification of objects, animals, and persons through radio waves.

2. How a security attack can succeed against EPC network architecture?Security attacks can target the different services of the EPC network architecture. Theymay succeed if weaknesses within the underlying technologies are not handled properly.The exchange of information between EPC tags and readers, for example, is carried out viawireless channels that do not possess basic security attributes such as authenticity, integrity,and availability. This situation allows attackers to misuse the RFID service of an EPC networkand perform unauthorized activities such as eavesdropping, rogue scanning, cloning, locationtracking, and tampering of data.

3. What is the major challenge for implementing new security features in EPC tags?The implementation of new security features in EPC tags faces several challenges. The mainone is the cost.

4. What is a threat?A threat is the objective of an attacker to violate security properties of a target system, such asauthenticity, integrity, and availability.

5. What is an attacker?An attacker is an agent that is exploiting a vulnerability of the targeted system to carry outthe threat. The exploitation of the vulnerability is defined as the attack.

6. Mention two major areas for research on security countermeasures for RFID technologies.Research on security countermeasures for RFID technologies can be divided into two cate-gories: (1) hardware-based security primitives for RFID tags and (2) software protocols usingthe hardware-based primitives.

7. What does ETSI stand for?European Telecommunications Standards Institute (ETSI)

8. When is a threat ranked as minor?A threat is ranked as minor when it is unlikely to happen or when its potential impact is low.Minor risk threats typically require no countermeasures.

9. Draw the diagram of Inventory Protocol of a Gen2 Tag.

“K11012_C003.tex” — page 61[#17] 20/8/2010 10:31


10. Define: Forward and Backward eavesdropping.We define forward eavesdropping as the passive collection of queries and commands sent fromreaders to tags; and backward eavesdropping as the passive collection of responses sent fromtags to readers.

Author’s BiographyJoaquin Garcia-Alfaro is a lecturer professor at the Computer Science and Multimedia Studiesof the Open University of Catalonia, Spain. He obtained a Bachelor, a Master’s, and a PhD (inComputer Science) from Autonomous University of Barcelona (Spain) and TELECOM Bretagne(France) in 2006. From 2007 to 2009, he was postdoctoral fellow at the Computer Science Depart-ment of Carleton University, Canada. Since 2009, he also collaborates as an associate researcherat TELECOM Bretagne, France. His research interests include a wide range of network securityproblems, with an emphasis on the management of security policies, analysis of vulnerabilities, andenforcement of countermeasures.

Michel Barbeau is a professor of computer science. He obtained a Bachelor, a Master’s, and a PhD,in computer science, from Université de Sherbrooke, Canada (1985), for undergraduate studies,and Université de Montréal, Canada (1987 and 1991), for graduate studies. From 1991 to 1999,he was a professor at Université de Sherbrooke. During the academic year 1998–1999, he was avisiting researcher at the University of Aizu, Japan. Since 2000, he works at Carleton University,Canada. Wireless communications has been his main research interest. He focuses his efforts onwireless security, vehicular communications, wireless access network management, ad hoc networks,and RFID.

Evangelos Kranakis is a professor of computer science. He obtained a BSc (in Mathematics) fromthe University of Athens (1973) and a PhD (in Mathematical Logic) from the University of Min-nesota, Minnesota (1980). From 1980 to 1982 he was at the Mathematics Department of PurdueUniversity, West Lafayette, Indiana, and from 1982 to 1983 at the Mathematisches Institut of theUniversity of Heidelberg, Germany, between 1983 and 1985 he served at the Computer ScienceDepartment of Yale University, New Haven, Connecticut, from August to December of 1985 at theComputer Science Department of the Universiteit van Amsterdam, the Netherlands and from 1986to 1991 at the Centrum voor Wiskunde en Informatica (CWI) in Amsterdam, the Netherlands.Since 1991, he works at Carleton University, Canada. He was director of the School of ComputerScience from 1994 to 2000. He received the Carleton Research Achievement award in 2000. Hebecame Carleton University Chancellor’s Professor in 2006. He has published in the analysis of algo-rithms, bioinformatics, communication and data (ad hoc and wireless) networks, computationaland combinatorial geometry, distributed computing, and network security.

References1. R. Want, RFID explained: A primer on radio frequency identification technologies, Synthesis Lectures

on Mobile and Pervasive Computing, Num. 1, 2006, Morgan & Claypool Publishers.2. G. Roussos, S. Duri, and W. Thompson. RFID meets the internet, IEEE Internet Computing. Special

Issue on RFID, 13, 105–114, 2009.3. EPCglobal Overview & Standards, Available from: http://www.epcglobalinc.org/standards/4. S. E. Sarma, Toward the 5 cent tag, White Paper, November 2001, Auto-ID Center.

“K11012_C003.tex” — page 62[#18] 20/8/2010 10:31


5. J. Sounderpandian, R. V. Boppana, S. Chalasani, and A. M. Madni, Models for cost–benefit analysis ofRFID implementations in retail stores, Systems Journal, IEEE, 1(2), 105–114, 2007.

6. ETSI, Methods and protocols for security; part 1: Threat analysis. ETSI-ts 102 165-1 v4.1.1, 2003.7. Y. Oren, Remote power analysis of RFID tags, Cryptology ePrint Archive, Report 2007/330, IACR, 2007.8. D. Han, T. Takagi, H. Kim, and K. Chung, New security problem in RFID systems tag killing, Lecture

Notes in Computer Science, M. L. Gavrilova, O. Gervasi, V. Kumar, C. J. K. Tan, D. Taniar, A. Lagana,Y. Mun, and H. Choo, eds. vol. 3982. Springer, 2006, pp. 375–384.

9. Minime and Mahajivana, RFID Zapper, 22nd Chaos Communication Congress (22C3), December 2005.10. M. Rieback, B. Crispo, and A. Tanenbaum, Is your cat infected with a computer virus?, in Pervasive

Computing and Communications, IEEE. Pisa, Italy: IEEE Computer Society Press, March 2006, pp.13–17.

11. P. Cole and D. Ranasinghe, eds. Networked RFID Systems and Lightweight Cryptography—Raising Barriersto Product Counterfeiting, 1st ed. Springer, 2008.

12. T. Lohmann, M. Schneider, and C. Ruland, Analysis of power constraints for cryptographic algorithmsin mid-cost RFID tags, in Smart Card Research and Advanced Applications, 7th IFIP WG 8.8/11.2 Inter-national Conference (CARDIS 2006), Lecture Notes in Computer Science, vol. 3928, Berlin/Heidelberg:Springer, April 2006, pp. 278–288.

13. M. Feldhofer and C. Rechberger, A case against currently used hash functions in RFID protocols,Workshop on RFID Security – RFIDSec 06, Ecrypt, Graz, Austria, July 2006, pp. 372–381.

14. S. Wolfram, Cryptography with cellular automata, in Advances in Cryptology, CRYPTO 85, Lecture Notesin Computer Sciences, vol. 218, New York, NY, USA: Springer-Verlag New York, Inc., 1986, pp. 429–432.

15. S. Sen, C. Shaw, D. R. Chowdhuri, N. Ganguly, and P. P. Chaudhuri, Cellular automata basedcryptosystem (CAC), in 4th International Conference on Information and Communications Security(ICICS’02). London, UK: Springer-Verlag, 2002, pp. 303–314.

16. P. Bardell, Analysis of cellular automata used as pseudorandom pattern generators, in International TestConference, 1990, Washington DC, pp. 762–768.

17. S. R. Blackburn, S. Murphy, and K. G. Paterson, Comments on theory and applications of cellularautomata in cryptography, IEEE Transactions on Software Engineering, 23(9), 637–638, 1997.

18. C. Meyer and W. Tuchman, Pseudo-random codes can be cracked, Electronic Design, 23, 1972.19. A. Menezes, P. Van Oorschot, and S. Vanstone, Handbook of Applied Cryptography. CRC Press, USA,

1997.20. J. Wolkerstorfer, Is elliptic-curve cryptography suitable to secure RFID tags? Workshop on RFID and

Lightweight Crypto, Ecrypt, Graz, July 2005.21. L. Batina, J. Guajardo, T. Kerins, N. Mentens, P. Tuyls, and I. Verbauwhede, An elliptic curve processor

suitable for RFID-tags, Cryptology ePrint Archive, Report 2006/227, IACR, 2006.22. M. Feldhofer, S. Dominikus, and J. Wolkerstorfer, Strong authentication for RFID systems using the

AES algorithm, in Workshop on Cryptographic Hardware and Embedded Systems, CHES 2004, LectureNotes in Computer Science, vol. 3156, IACR. Boston, MA, USA: Springer, August 2004, pp. 357–370.

23. J. Daemen and V. Rijmen, The Design of Rijndael: AES—the Advanced Encryption Standard. Springer,2002.

24. S. Mangard, M. Aigner, and S. Dominikus, A highly regular and scalable AES hardware architecture,IEEE Transactions on Computers, 52(4), 483–491, 2003.

25. I. Verbauwhede, P. Schaumont, and H. Kuo, Design and performance testing of a 2.29-GB/s Rijndaelprocessor, IEEE Journal of Solid-State Circuits, 38(3), 569–572, 2003.

26. A. Juels, RFID security and privacy: A research survey, IEEE Journal on Selected Areas in Communication,24(2), 381–394, 2006.

27. P. Israsena, Securing ubiquitous and low-cost RFID using tiny encryption algorithm, in InternationalSymposium onWireless Pervasive Computing, IEEE. Phuket, Thailand: IEEE Press, January 2006, pp. 1–4.

28. F. Mace, F.-X. Standaert, and J.-J. Quisquater, Asic implementations of the block cipher sea forconstrained applications, in Conference on RFID Security, Malaga, Spain, July 2007, pp. 103–114.

29. D. J. Wheeler and R. M. Needham, TEA, a tiny encryption algorithm, in Fast Software Encryption:Second International Workshop (FSE 1994), Leuven, Belgium, December, Lecture Notes in ComputerScience, vol. 1008. Springer, Berlin/Heidelberg, 1995, pp. 363–366.

“K11012_C003.tex” — page 63[#19] 20/8/2010 10:31


30. K. Fishkin, S. Roy, and B. Jiang, Some methods for privacy in RFID communication, in EuropeanWorkshop on Security in Ad-hoc and Sensor Networks, ESAS 2004, Lecture Notes in Computer Science, vol.3313. Heidelberg, Germany: Springer-Verlag, August 2005, pp. 42–53.

31. G. Hancke, Noisy carrier modulation for HF RFID, in First International EURASIP Workshop on RFIDTechnology, Vienna, Austria, September 2007, pp. 63–66.

32. A. Solanas, J. Domingo-Ferrer, A. Martinez-Balleste, and V. Daza, A distributed architecture for scalableprivate RFID tag identification, Computer Networks, 51(9), 2268–2279, 2007.

33. D. Molnar, A. Soppera, and D. Wagner, A scalable, delegatable pseudonym protocol enabling ownershiptransfer of RFID tags, in B. Preneel and S. Tavares, eds. Selected Areas in Cryptography, SAC 2005, LectureNotes in Computer Science, vol. 3897. Kingston, Canada: Springer, August 2005, pp. 276–290.

34. J. Hall, M. Barbeau, E. and Kranakis, Enhancing intrusion detection in wireless networks using radiofrequency fingerprinting, Communications, Internet, and Information Technology, 2004, 201–206.

35. J. Hall, Detection of rogue devices in Wireless Networks, PhD dissertation, Carleton University,2006.

36. G. Avoine and P. Oechslin, RFID traceability: A multilayer problem, in Financial Cryptography 2005,Lecture Notes in Computer Science, vol. 3570, IFCA. Roseau, The Commonwealth Of Dominica:Springer-Verlag, February–March 2005, pp. 125–140.

37. R. Pappu, Physical one-way functions, PhD dissertation, MIT, 2001.38. B. Gassend, Physical random functions, Master’s thesis, MIT, 2003.39. B. Gassend, D. Clarke, M. van Dijk, and S. Devadas, Silicon physical random functions, in 9th

ACM Conference on Computer and Communications Security. New York, NY, USA: ACM, 2002,pp. 148–160.

40. D. Lim, J. Lee, B. Gassend, G. Suh, M. van Dijk, and S. Devadas, Extracting secret keys from integratedcircuits, IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 13(10), 1200–1205, 2005.

41. B. Skoric and P. Tuyls, Secret Key Generation from Classical Physics, Philips Research Book Series, Septem-ber 2005.

42. D. Holcomb, W. Burleson, and K. Fu, Initial SRAM state as a fingerprint and source of true randomnumbers for RFID tags, in Third International Conference on RFID Security, RFIDSec 2007, Malaga,Spain, 2007, pp. 31–42.

43. D. Ranasinghe, D. Engels, and P. Cole, Security and privacy solutions for low cost RFID Systems,in 2004 Intelligent Sensors, Sensor Networks & Information Processing Conference, Melbourne, Australia,2004, pp. 337–342.

44. L. Bolotnyy and G. Robins, Physically unclonable function-based security and privacy in RFID systems,in International Conference on Pervasive Computing and Communications—PerCom 2007, IEEE. NewYork, USA: IEEE Computer Society Press, March 2007, pp. 211–220.

45. K. Takaragi, M. Usami, R. Imura, R. Itsuki, and T. Satoh, An ultra small individual recognition securitychip, IEEE Micro, 21(6), 43–49, 2001.

46. A. Willig, M. Kubisch, C. Hoene, and A. Wolisz, Measurements of a wireless link in an industrialenvironment using an IEEE 802.11-compliant physical layer, IEEE Transactions on Industrial Electronics,49(6), 1265–1282, 2002.

47. A. Juels and R. Pappu, Squealing euros: Privacy protection in RFID-enabled banknotes, in R. N.Wright, ed., Financial Cryptography 2003, Lecture Notes in Computer Science, vol. 2742, IFCA. LeGosier, Guadeloupe, French West Indies: Springer, January 2003, pp. 103–121.

48. X. Zhang and B. King, Integrity improvements to an RFID privacy protection protocol foranti-counterfeiting, in J. Zhou, J. Lopez, R. Deng, and F. Bao, eds. Information Security Conference,ISC 2005, Lecture Notes in Computer Science, vol. 3650. Singapore: Springer, September 2005, pp.474–481.

49. S. Kinosita, F. Hoshino, T. Komuro, A. Fujimura, and M. Ohkubo, Non identifiableanonymous-ID scheme for RFID privacy protection, in Japanese. English description as part ofhttp://www.autoidlabs.com/whitepaper/KEI-AUTOID-WH004.pdf, 2003.

50. A. Juels, Minimalist cryptography for low-cost RFID tags, in C. Blundo and S. Cimato, eds. InternationalConference on Security in Communication Networks, SCN 2004, Lecture Notes in Computer Science, vol.3352. Amalfi, Italia: Springer, September 2004, pp. 149–164.

“K11012_C003.tex” — page 64[#20] 20/8/2010 10:31


51. S. Weis, S. Sarma, R. Rivest, and D. Engels, Security and privacy aspects of low-cost radio frequencyidentification systems, in D. Hutter, G. Müller, W. Stephan, and M. Ullmann, eds., InternationalConference on Security in Pervasive Computing, SPC 2003, Lecture Notes in Computer Science, vol. 2802.Boppard, Germany: Springer, March 2003, pp. 454–469.

52. D. Henrici and P. Müller, Hash-based enhancement of location privacy for radio-frequency identificationdevices using varying identifiers, in International Workshop on Pervasive Computing and CommunicationSecurity—PerSec 2004, IEEE Computer Society, March 2004, pp. 149–153.

53. G. Avoine and P. Oechslin, A scalable and provably secure hash based RFID protocol, in InternationalWorkshop on Pervasive Computing and Communication Security—PerSec 2005, IEEE Computer SocietyPress, March 2005, Kauai Island, Hawaii, pp. 110–114.

54. S.-M. Lee, Y. J. Hwang, D. H. Lee, and J. I. L. Lim, Efficient authentication for low-cost RFID systems,in O. Gervasi, M. Gavrilova, V. Kumar, A. Lagana‘a, H. P. Lee, Y. Mun, D. Taniar, and C. J. K. Tan,eds. International Conference on Computational Science and its Applications, ICCSA 2005, Lecture Notesin Computer Science, vol. 3480. Singapore: Springer, May 2005, pp. 619–627.

55. E. Y. Choi, S. M. Lee, and D. H. Lee, Efficient RFID authentication protocol for ubiquitous computingenvironment, in T. Enokido, L. Yan, B. Xiao, D. Kim, Y. Dai, and L. Yang, eds., International Workshopon Security in Ubiquitous Computing Systems, SECUBIQ 2005, Lecture Notes in Computer Science, vol.3823. Nagasaki, Japan: Springer, December 2005, pp. 945–954.

56. S. Lee, T. Asano, and K. Kim, RFID mutual authentication scheme based on synchronized secretinformation, in Symposium on Cryptography and Information Security, Hiroshima, Japan, January 2006.

57. G. Tsudik, YA-TRAP: Yet another trivial RFID authentication protocol, in International Conference onPervasive Computing and Communications, PerCom 2006, IEEE. Pisa, Italy: IEEE Computer SocietyPress, March 2006, pp. 640–643.

58. C. Chatmon, T. van Le, and M. Burmester, Secure anonymous RFID authentication protocols, FloridaState University, Department of Computer Science,Tallahassee, FL, USA,Technical ReportTR-060112,2006.

59. A. Juels, R. Pappu, and B. Parno, Unidirectional key distribution across time and space with applicationsto RFID security, in USENIX Security Symposium. San Jose, CA: USENIX, July 2008, pp. 75–90.

60. D. Ranasinghe, D. Engels, and P. Cole, Low-cost RFID systems: Confronting security and privacy, inAuto-ID Labs Research Workshop, Zurich, Switzerland, September 2004, pp. 54–77.

61. P. Tuyls and L. Batina, RFID-tags for anti-counterfeiting, in Topics in Cryptology, CT-RSA 2006, TheCryptographers’ Track at the RSA Conference 2006, Lecture Notes in Computer Science, USA: Springer,February 2006, pp. 115–131.

62. A. Juels and S. Weis, Authenticating pervasive devices with human protocols, in Advances in Cryptology,CRYPTO 2005, Lecture Notes in Computer Science, vol. 3126, IACR. Santa Barbara, CA, USA: Springer,August 2005, pp. 293–308.

63. D. Molnar and D. Wagner, Privacy and security in library RFID: Issues, practices, and architectures, inConference on Computer and Communications Security—ACM CCS, USA: ACM Press, October 2004,pp. 210–219.

64. J. Bringer, H. Chabanne, and T. Icart, Improved privacy of the tree-based hash protocols using physicallyunclonable function, Cryptology ePrint Archive, Report 2007/294, 2007.

handling securitythreats to the rfid system of epc networksgarcia_a/web/papers/... · used in rfid...

Documents